The rule and the evidence: A study interviewed 12 people in Italy about disinformation

### THE ACT

This article is not a resolution, a ruling, an audit, a bill or any other act in force: it is a qualitative study of how Italian institutional actors perceive disinformation, accepted for publication but not yet copyedited or typeset, and it names no rule, no enforcement body, no obligation on any person and no deadline¹.

The only thing in it that resembles a measure in force is a European Commission definition it quotes: disinformation is information that is "created, presented and diffused for economic gain or to intentionally deceive the public" and that "can cause public harm"¹.

The research behind it was commissioned to Antonella Coco by the Cabinet Office of the Italian Minister of Defence in 2023, and the article does not state who evaluated it or any conflicts of interest¹.

### THE CASE IT ANSWERS

No real instance was found this week.

### WHAT WAS MEASURED

The study measured nothing in the reader's units: it is a qualitative analysis whose empirical base is 12 semi-structured interviews, of which 3 were exploratory interviews with scholars, 6 were with institutional actors involved in observing or countering disinformation, and 3 were with journalists, communicators and analysts¹.

The article reports no sample of voters, no exposure rate, no persuasion effect, no targeting figure and no turnout figure, and it says the research design distinguishes the institutional level of identifying the risk of disinformation from the dynamics of content production and from their effects on public opinion¹.

It has never been measured in an election in this study, because the study measured no election: it asked 12 people in politics, defence, communication and research what they think the threat is¹.

The one number the article offers about the size of the problem is that disinformation is recognised as a public problem by 54% of the countries examined, which have implemented specific policies and institutional actions¹ — a little more than half of the countries looked at, and nothing about any single vote.

### THE OTHER REGION'S ANSWER TO THE SAME THING

We found none this week.

### THE LIBRARY'S READING

Here is how we read it. A false story about a candidate has always done the same work: it does not need to be believed to do damage, only repeated, because once a charge is in the air the accused carries the smell of it. What is new is the speed and the reach, not the trick.

What a public can learn to check is limited and specific: whether a source is named, whether it is independent, whether it answers for what it says, whether the same claim appears anywhere outside the circle that already agrees with it. What it cannot check alone is whether the thing it is looking at was made by a machine, and that is the part that now decides what people forward to each other.

So expect this in homes like yours: the argument after a vote will not be about a fact but about whom to believe, and each side will bring its own account of the same ballot and treat the other's as invented. You would know we are wrong if people who cannot verify a claim still sorted true from false by going to the electoral authority's own published record, and if their trust in the count followed the audits rather than the noise.

What you can do with it: before you vote, ask which body actually vouches for the claim in front of you, and whether that body publishes its evidence where you can open it. When a rumour about the count goes around, point people to the authority's own audit rather than to whoever repeated it loudest — and ask what single document both sides would accept as settling it, then notice if the answer is nothing.

### WHAT IS NOT KNOWN

The article does report one specific deepfake: the BuzzFeed News video of April 2018 in which Barack Obama appears to say things he never said, made with the actor Jordan Peele, and used to show how easily a false message can look real and spread².

The article discusses the possible role of artificial intelligence in countering disinformation, and it says the study did not evaluate any intervention or weigh any technology's overall benefit¹.

It is not known whether what the 12 Italian interviewees describe holds in your country, because the study's subjects are Italian institutional actors in politics, defence, communication and research¹.

What the record is silent on is whether any register, count or result was affected, and how many voters were ever reached: the study records perceptions of a threat, not its occurrence, and it describes disinformation as a factor that erodes institutional trust¹.

The one thing the interviewees do say about the mechanism is that the aim is not to plant doubt about a particular claim or person but to leave behind distrust itself³ — and that the distrust is both an effect of disinformation and one of the attackers' main objectives⁴.

The article also says the threat is described as permanent and continuous, without a real distinction between peacetime and wartime⁵, and that its cross-border character makes it very hard to say where a message came from and therefore very hard to counter⁶.

On what to do instead, the article says fact-checking, debunking and pre-bunking appear largely reactive and insufficient¹, and that the answer preferred by those interviewed is not security measures but an approach built on shared responsibility, schooling, media literacy and critical thinking¹.

And here is the last thing worth carrying to the polling station: the study tells you what people who watch this threat believe about it, not what happened to a single vote. So when someone tells you the count was stolen, ask them for the one document they would accept as proof either way — and if they have none, you already know what you are standing in.